← Feed

Posts

• View post

  Managing an AWS Organization With AI Generated Code A new take on the AWS Control Tower concept https://teriradichel.substack.com/p/managing-an-aws-organization-with

• View post

  Multi-threaded programming is hard. I would say AI 🤖 models are bad at it based on how long this took and I had to figure out how to solve some of the core blockers myself, but honestly the model built the base parallel processor to deploy things in multiple accounts at once way faster than I would. It also helped me understand the ins and outs of how it worked - but it could not solve the problem with access to its own code and read permissions to the other projects. I had a heck of a time ge...

• View post

  So exhausting. Finally. AI is my nemesis and my addiction and downfall and my tool to conquer all at the things all the same time. Really killed me this week like three times when I was almost done did something really wonky. I keep finding ways to reign it in. Will be easier when this is complete. You get to a point where it will take longer to go back so you just have to get it done. I’m so behind and barely slept this week but finally got this thing working. The main thing right now is to...

• View post

  I crack myself up. I responded to an AI 🤖 agent nonsense comment with a plausible sounding AI generated nonsensical answer and it is replying telling me how great that is and having a completely nonsensical conversation about nothingness.

• View post

  Why am I seeing AirPods tracking notifications on my phone when I do not use AirPods. Anyone?

• View post

  Why am I seeing AirPods tracking notifications on my phone when I do not use AirPods. Anyone?

• View post

  Chinese Threat Actor Uses Leaked DarkSword Kit to Deploy GHOSTBLADE on iOS …threat actor running more than 100 web properties, most of which are fake Amazon Web Services (AWS) sign-in pages on a domain that also hosts the exploit toolkit. https://thehackernews.com/2026/08/chinese-threat-actor-uses-leaked.html

• View post

  5 days ago I wrote this on X: So let’s just guess that the package manager was JFrog. If you are running JFrog keep a close eye on your network traffic. Unfortunately anything with Internet access, if compromised, can become a proxy. Like DNS servers. Not an easy problem to solve. ~~ I was right. JFrog was the third party package manager the agent escaped in the OpenAI / Hugging Face incident. Now about everything else…I hope people are starting to understand how network security is one of y...

• View post

  SSH Key on a Yubikey ~ Almost Protecting credentials from rogue AI agents and malware on your laptop and how Apple makes it difficult https://teriradichel.substack.com/p/ssh-key-on-a-yubikey-almost

• View post

  SSH Key on a Yubikey ~ Almost Protecting credentials from rogue AI agents and malware on your laptop and how Apple makes it difficult

• View post

  Revisiting the Bastion Host In Light of AI Agent Escapes A bastion host adds a point of inspection for potential rogue traffic https://teriradichel.substack.com/p/revisiting-the-bastion-host-in-light

• View post

  Revisiting the Bastion Host In Light of AI Agent Escapes A bastion host adds a point of inspection for potential rogue traffic https://teriradichel.substack.com/p/revisiting-the-bastion-host-in-light

• View post

  Using gpt terra trying to delete time files just like Anthropic did. Completely messing up multithreaded architecture. Accidentally deleted files and tried to restore from five day old backups. Chinese characters appeared on screen at some point. In no way proves this is Chinese issue. Could be random or someone trying to make it look like a Chinese hacker. I have no idea. I couldn’t get enough info. But the model tried to tell me it wasn’t Chinese. Google AI says it is some Chinese gambling s...

• View post

  This is so awesome and a great reason to use Kiro CLI (@kirodotdev Kiro.dev) Now you can select from Anthropic and OpenAI models in Kiro. Just type /model and select the model to use. Why this is so cool… I have a custom agent framework and run different agents in different terminal windows. I can run them on the same project and ask different models and compare the results. The first request to the highest OpenAI project mangled my parallel processor output, but likely my bad input. I fix...

• View post

  RE: https://infosec.exchange/@teriradichel/116925067367520094 Can you build real world software with AI? Yes. But it’s not like you just let an agent do it all. It’s a lot harder than that. This took about 4 months, long hours, a few hiatuses. And it’s complicated, multithreaded, parallel processing with a complex underlying updatable architecture. But I’ve been trying to get something like this done for the past 5 years without AI and AI made it possible to get it to a working state. I wouldn’...

• View post

  RE: https://infosec.exchange/@teriradichel/116839536008038553 A few core things that would help AI 🤖 models burn less tokens. 💰 Yesterday was all manual. The agents were just pounding on some things and not fixing certain bugs so I turned it off and dug into the code. So many errors were hidden by not pushing the bugs out of the subshell to the surface or not writing a message at all after an error occurs < and that is how AI models “fix” a lot of bugs. They don’t really fix the underlying...

• View post

  🥲 Today felt like a waste of time mostly. The models were so slow and made so many mistakes I had to walk them through every little thing. They reverted bug fixes I made manually. Another $200 burned. I had a few tricky problems they couldn’t figure out themselves and even though I did a lot of the legwork today still ran me out of credits. What happened right before it times out was one of the agents deleted some diagram recursive xml code I had painstakingly walked it through creating and...

• View post

  More to come on the blog when time allows. Hope the agents have all the bugs fixed by the time I’m back from running errands. Blog: ✍️👩‍💻🤖🔒 https://teriradichel.substack.com

• View post

  1123 bugs fixed this week. 🐞🐛🐜🦟🪲🪳🕷️ By My AI agents. 🤖🤖🤖🤖🤖🤖🤖🤖🤖 But Kiro/Anthropic going so S-L-O-W 🥱 right now. The bug counts will not move as fast as a result. Opus 4.8. Some bugs are actually architectural changes or features but mostly bugs. I gave up on clean code. It is ok but not as perfect as I would like. Too much duplication though I have strict rules and tests for critical components. I figured out I was wasting tokens having agents write the bugs except for some bulk updates....

• View post

  The loudest and most reprimanding voices are not always the most right voices. I’m just over here writing about what I experience when I actually use AI models and reposting other’s empirical evidence, not hot takes, opinions, or other such noise. Choose wisely. Do your own research.

• View post

  Most cybersecurity workers have been told to conceal a breach report finds. -Cybersecurity Dive https://www.cybersecuritydive.com/news/data-breach-coverups-ai-bitdefender/824331/

• View post

  It’s July 1st and I’ve used up all the tokens in a $200 plan this month already. So that’s how it’s going. Fable came out. Opus is dumber and slower. Happens every time a new model comes out even when it’s not available yet. Though new version of sonnet came to AWS. But I’m not using sonnet. At one point I asked the model which model it was. Twice it said it inferred that it was Sonnet. But then it said it was guessing.

• View post

  Managing an AWS Organization With AI Generated Code A new take on the AWS Control Tower concept https://teriradichel.substack.com/p/managing-an-aws-organization-with

• View post

  ⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️ Yesterday the 🤖 model randomly tried to source a (thankfully) non-existent delete-all-accounts file in a file that was sourced by a file that is only supposed to list all the accounts in an AWS OU. I never told it to source either of those rogue files. Let that sink in. ⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️

• View post

  I have to take care of a family member. Headed out to help her for I don’t know how long but hope to be working on AI again soon. Trying to enjoy the scenery along the way waiting for her to get back home.

• View post

  Note: As part of the analyzed intrusions, public-facing applications and valid accounts were abused for initial access. The state-sponsored hackers targeted Ivanti, Cisco, Fortinet, VMware, and Palo Alto Networks appliances, as well as Apache Struts and other web-facing platforms. - Are these 100% American products? Buying American doesn’t mean we are safe. - The tactics used here show exactly how stealthy malware can be. A shell triggered by a particular byte sequence? Something that puts its...

• View post

  Latest scam. Interesting that when I used Apple cleanup to remove the ticket number it didn’t remove it but instead characters from another Arabic looking language showed up instead. This came as an attachment. Took screenshot. I did the cleanup like 5 times and then chars gone.

• View post

  How I Use AI for Penetration Testing Speaking at the Computer History Museum in Mountain View, CA April 10, 2026 https://teriradichel.substack.com/p/how-i-use-ai-for-penetration-testing

• View post

  Iran has rapidly developed advanced cyber capabilities, evolving from information gathering to conducting destructive, state-linked attacks against critical infrastructure in the U.S., Israel, and the Gulf states. https://share.google/aimode/50qKfH5TPLWSJFVOB

• View post

  I was just listening to an interview on the radio with a person who worked at a hospital. 1. Your cyber insurance makes you a target. They know how much you can pay. 2. Don’t use your backups until you have eliminated the attacker or they will encrypt your backups too. 3. Pull the plug until you figure that out and cut them off (except critical patient machines). The hospital in story I was listening to had people running across the hospital when faxes were overused and started smoking....